Autonomous System Control in Unknown Operating Conditions
Autonomous systems have become an interconnected part of everyday life with the recent increases in computational power available for both onboard computers and offline data processing. Two main research communities, Optimal Control and Reinforcement Learning stand out in the field of autonomous systems, each with a vastly different perspective on the control problem. While model-based controllers offer stability guarantees and are used in nearly all real-world systems, they require a model of the system and operating environment. The training of learning- based controllers is currently mostly limited to simulators, which also require a model of the system and operating environment. It is not possible to model every possible operating scenario an autonomous system can encounter in the real world at design time and currently, no control methods exist for such scenarios. In this seminar, we present a hybrid control framework, comprised of a learning-based supervisory controller and a set of model-based low-level controllers, that can improve a system’s robustness to unknown operating conditions.
Yves Sohege, affiliated to the Insight Centre for Data Analytics, University College Cork (UCC), Ireland, is a 26-year-old PhD student at UCC. He have been working remotely from the Netherlands for the past two years. He completed his BSc in Computer Science in UCC in 2016 during which he was nominated for Graduate of the Year and received the Quercus Scholarship for Academic Performance. His research interests are focused around Reinforcement Learning and fault-tolerant control, specifically focusing on unknown operating conditions and quadcopter control.
Wednesday, April 7, 2021, at 14:00
ID meeting: 999 5456 5468 (no password needed)
CSCS introduction for IDSIA researchers
Mario Valle, Henrique de Almeida Mendonça, Rafael Sarmiento Perez, Guilherme Peretti-Pezzi
The goal of this talk is to briefly introduce the CSCS infrastructure, including the allocation mechanisms and services that could potentially be relevant to researchers from IDSIA. We will also be available for a Q&A session after the presentation.
- Mario started more than 30 years ago to work as a computer scientist playing various roles in various fields. Since 2003 is at CSCS working as scientific visualization expert with chemists and crystallographers to move afterwards to a more general data scientist job. Currently he is in charge of the Persistent Identifier service at CSCS.
- Henrique has more than 10 years of experience in distributed and high performance systems. He has worked in the industry in projects involving classical and geometric computer vision and deep learning, with extensive experience in C++, CUDA, PyTorch and TensorFlow. He is also quite active in the Kaggle community, as well as other ML competitions.
- Rafael is part of the Compute and Data Services Support group of the User Engagement and Support unit at CSCS. He obtained his PhD in Physics from the University of Lyon I in 2015. There, he worked on ab initio calculations and prediction of crystal structures. He then moved to the University of Basel for a postdoctoral fellowship on machine learning applied to the prediction of chemical properties of molecular systems. In 2017 he started his current position at CSCS.
- Guilherme is leading the Compute and Data Service Support group at CSCS. He has a PhD degree in Computer Science and 10+ years of experience in the field of HPC. He worked for both academia, by performing research on parallel programming models, and for industry, with the development of numerical software for computational fluid dynamics simulations.
Thursday, March 18th, 2021, at 11:30
ID meeting: 925 2738 8637 (no password needed)